Making AI Experience Many Trials to Achieve High-level Behavior

ReNom | ReNomRL
Deep reinforcement learning is a combination of deep learning and reinforcement learning. The AI learns to choose optimal action through multiple trials, by experiencing “state” that it is in, “action” taken from
the current state, “state” changed by action, and “reward” it gets for an action. Cases where machine winning against a human player in the game GO, AI recording high scores in games, or machine automatically controlling manufacturing process, are representation of deep reinforcement learning being utilized.


About ReNomRL

ReNom RL is a module for creating deep reinforcement learning models. The addition of the ReNom RL Python API to the Machine Learning / Deep Learning AI development platform ReNom allows users to quickly create, train and evaluate reinforcement learning models and select the most effective model for the task at hand.


Learning Experience Through Action and Reward

Labeled dataset is used in supervised learning, but in reinforcement
learning, rewards received by the action AI takes from the current state
are used to learn optimal behavior. As if a human being learns how to
play naturally with repetitive experience, AI also learns by accumulating


Deep Reinforcement Learning Algorithm

ReNom RL has built-in classes for deep reinforcement learning
algorithms, such as DQN (Deep Q Network), DDPG (Deep Deterministic
Policy Gradient), RDPG (Recurrent Deterministic Policy Gradient), and
A3C (Asynchronous Advantage Actor-Critic), and can start the required
learning process by just defining the reward function. We are also
actively engaged in research activities with universities as new
reinforcement learning technology are developed.



In, we provide tutorials using OPEN AI gym for deep
reinforcement learning and as well as guide viewers on how to
use reinforcement learning.


Page top